Transaction Reporting: Filter vs. Feed
Navigating the operational shift from suspicious activity monitoring to mandatory threshold reporting
The Two Philosophies of Financial Intelligence
To the uninitiated, ‘Anti-Money Laundering’ (AML) looks like a single global standard. You verify the user (KYC), you monitor the transaction, and you report the bad guys.
But for the fintech innovator expanding internationally, this assumption is a dangerous architectural blind spot. In reality, the world is split into two distinct surveillance philosophies: The Filter and The Feed.
Mistaking one for the other doesn’t just annoy regulators; it breaks your operational scalability.
The Filter: The US & UK Model
In jurisdictions like the USA (FinCEN) and UK (NCA), the regulator essentially deputises you as an investigator. They do not want to see every transaction. They want you to apply a ‘filter’—your own risk logic—and only send them what looks suspicious (via a SAR).
In this model, your engineering challenge is anomaly detection. You are building a needle-finder. If a user wires $50,000, the regulator says: ‘Keep the record, but don’t bother us unless it looks wrong.’
The Feed: The Canada & Australia Model
Then there is the ‘feedf model, championed by Canada (FINTRAC) and Australia (AUSTRAC). These regulators don’t just want the needles; they want the entire haystack.
Canada’s Large Virtual Currency Transaction Report (LVCTR) is the prime example. It is a mandatory, automatic firehose. If a user receives over $10,000 CAD in crypto—regardless of how clean, legitimate, or routine the transaction is—you must send a structured file to Ottawa.
In this model, your engineering challenge is aggregation. The regulator’s ‘24-Hour Rule’ means you cannot just look at single transactions; you must maintain a rolling state of every user’s activity. If a user receives three payments of $3,500 within 24 hours, you have hit the $10,500 trigger. If your ledger doesn’t aggregate that instantly, you are in breach.
The Strategic Divergence
For the Product Lead, this creates a bifurcation in your roadmap:
The ‘Intelligence’ Stack (US/UK/EU): You need expensive ML models and human analysts to reduce false positives, because over-reporting is seen as ‘defensive filing’ and frowned upon.
The ‘Counter’ Stack (Canada/Australia): You need rigid, deterministic logic. There is no judgement call. If
Sum(Transactions_24h) >= 10,000, thenReport().
The Future: The Convergence of Zero
While the US sticks to the ‘filter,’ Europe is slowly moving toward the ‘feed’—but for tax, not crime. The upcoming DAC8 framework (2026) effectively creates a near-zero threshold for crypto reporting to tax authorities.
The trend is clear: the days of ‘record keeping’ (holding data just in case) are ending. The era of ‘mandatory reporting’ (streaming data by default) is beginning.
Actionable Horizon Scanning
Reporting triggers are hard-coded constraints, not policy guidelines. Pericls maps these mandatory reporting thresholds—distinguishing between ‘suspicious activity’ triggers and ‘automatic threshold’ triggers—directly to your jurisdiction profile, ensuring your engineering team builds the right pipeline for the right regulator.
The Pericls Team